RFC 1661, "The Point-to-Point Protocol (PPP)", July 1994

Note: This RFC has been updated by RFC 2153

Errata ID: 543
Status: Verified
Type: Editorial
Publication Format(s) : TEXT
Reported By: Stuart Cheshire
Date Reported: 2006-09-27
Verifier Name: Brian Haberman
Date Verified: 2012-04-26

Section 5.4 says:

On reception of a Configure-Reject, the Identifier field MUST
match that of the last transmitted Configure-Request.
Additionally, the Configuration Options in a Configure-Reject MUST
be a proper subset of those in the last transmitted Configure-
Request.  Invalid packets are silently discarded.

It should say:

On reception of a Configure-Reject, the Identifier field MUST
match that of the last transmitted Configure-Request.
Additionally, the Configuration Options in a Configure-Reject MUST
be a subset of those in the last transmitted Configure-
Request.  Invalid packets are silently discarded.

Notes:

The word "proper" should be deleted. (I discussed this a while ago with
the author, Bill Simpson, and he agreed.)

If A is a subset of B, then set A contains only elements that are also in
set B, up to and including all the elements of B (in which case A == B).

If A is a PROPER subset of B, then A contains only elements that are in
B, up to BUT NOT including all the elements of B.

In this case, the word "proper" is just a mistake, perhaps added because
someone thought it made the sentence sound better, without realizing that
the mathematical term "proper subset" has a specific precise meaning,
which is the wrong meaning in this case. As written in the RFC, the
sentence states that if a Configure-Request packet has n Configuration
Options in it, ALL of which are not recognizable or not acceptable, then
when the implementation returns its Configure-Reject packet, it's only
allowed to indicate that up to n-1 options were rejected, when in truth
ALL were rejected. Removing the word "proper" removes this unintended and
incorrect limitation.

Report New Errata